# MeteoSignal: results Independent verification of open weather forecast data: ECMWF's AI model AIFS against IFS and GFS. Data version 16 November 2026, run 20261116T194902Z. All results are provisional: no AIFS version has been scored over a full year. Each check has its own places, days and caveats, so numbers are not to be compared between checks. "Tested" means the difference passed a block bootstrap and a Holm-corrected test on the whole sample; the method is at https://meteosignal.com/method/. ## Germany 30 DWD stations in Germany. Scored against: Hourly station observations (DWD open data). AIFS 2: 174 days, 23 May 2026 to 12 November 2026. Other AI models: 68 days. The main result, with tested verdicts. AIFS 2, raw forecasts: AIFS's RMSE was lower in 17 of 28 tested comparisons, higher in 0, and could not be separated in 11. Each value is AIFS's RMSE minus the other model's, in per cent of the other model's. Minus is a lower AIFS error (better). | Variable | Against | Day 1 | Day 3 | Day 5 | Day 7 | Day 10 | Days | |---|---|---|---|---|---|---|---| | 2 m temperature | IFS | -10% (lower, tested) | -13% (lower, tested) | -13% (lower, tested) | -9% (not separated) | -11% (not separated) | 174 | | 2 m temperature | GFS | -29% (lower, tested) | -29% (lower, tested) | -24% (lower, tested) | -20% (not separated) | -15% (not separated) | 174 | | 10 m wind | IFS | -1% (not separated) | -3% (not separated) | -7% (lower, tested) | -5% (not separated) | 0% (not separated) | 174 | | 10 m wind | GFS | -4% (not separated) | -8% (lower, tested) | -10% (lower, tested) | -10% (lower, tested) | -3% (not separated) | 174 | | Solar energy | IFS | -15% (lower, tested) | -16% (lower, tested) | -13% (lower, tested) | not scored | not scored | 174 | | Solar energy | GFS | -23% (lower, tested) | -29% (lower, tested) | -25% (lower, tested) | -19% (lower, tested) | -13% (not separated) | 174 | Page: https://meteosignal.com/germany/ ## Europe 107 stations in six countries, Germany included. Scored against: Measurements from the national services of Germany, Austria, the Netherlands, Denmark, Switzerland and Finland. AIFS 2: 174 days, 23 May 2026 to 12 November 2026 for temperature and wind; solar energy 173 days. One pooled set with its own tests. A further 61 stations in 11 countries, from NOAA's coarser hourly reports, are scored as a set of their own. AIFS 2, raw forecasts: AIFS's RMSE was lower in 24 of 28 tested comparisons, higher in 0, and could not be separated in 4. Each value is AIFS's RMSE minus the other model's, in per cent of the other model's. Minus is a lower AIFS error (better). | Variable | Against | Day 1 | Day 3 | Day 5 | Day 7 | Day 10 | Days | |---|---|---|---|---|---|---|---| | 2 m temperature | IFS | -10% (lower, tested) | -12% (lower, tested) | -13% (lower, tested) | -9% (not separated) | -9% (not separated) | 174 | | 2 m temperature | GFS | -28% (lower, tested) | -27% (lower, tested) | -24% (lower, tested) | -20% (lower, tested) | -14% (lower, tested) | 174 | | 10 m wind | IFS | -3% (lower, tested) | -5% (lower, tested) | -8% (lower, tested) | -6% (not separated) | -1% (not separated) | 174 | | 10 m wind | GFS | -15% (lower, tested) | -16% (lower, tested) | -16% (lower, tested) | -15% (lower, tested) | -8% (lower, tested) | 174 | | Solar energy | IFS | -16% (lower, tested) | -17% (lower, tested) | -15% (lower, tested) | not scored | not scored | 173 | | Solar energy | GFS | -23% (lower, tested) | -27% (lower, tested) | -24% (lower, tested) | -21% (lower, tested) | -14% (lower, tested) | 173 | Page: https://meteosignal.com/europe/ ## Cabauw tower Cabauw tower, the Netherlands. Scored against: Measurements on KNMI's Cabauw tower. AIFS 2: 174 days, 23 May 2026 to 12 November 2026 for wind; temperature 172 days. A case study at one site, never pooled with the stations. Some months use measurements KNMI has not yet verified. AIFS 2, raw forecasts: AIFS's RMSE was lower in 7 of 20 tested comparisons, higher in 0, and could not be separated in 13. Each value is AIFS's RMSE minus the other model's, in per cent of the other model's. Minus is a lower AIFS error (better). | Variable | Against | Day 1 | Day 3 | Day 5 | Day 7 | Day 10 | Days | |---|---|---|---|---|---|---|---| | 2 m temperature | IFS | -12% (not separated) | -14% (not separated) | -16% (not separated) | -13% (not separated) | -6% (not separated) | 172 | | 2 m temperature | GFS | -45% (lower, tested) | -39% (lower, tested) | -32% (lower, tested) | -29% (not separated) | -16% (not separated) | 172 | | 100 m wind | IFS | -10% (not separated) | -5% (not separated) | -10% (not separated) | -8% (not separated) | -6% (not separated) | 174 | | 100 m wind | GFS | -22% (lower, tested) | -23% (lower, tested) | -22% (lower, tested) | -15% (not separated) | -19% (lower, tested) | 174 | Page: https://meteosignal.com/cabauw/ ## ERA5 over Europe Europe, against ERA5. Scored against: ERA5 reanalysis on a 0.25° grid, 35 to 72°N and 12.5°W to 40°E. AIFS 2: 171 days, 23 May 2026 to 9 November 2026. Other AI models: 108 days. ERA5 is a model product and AIFS was trained mostly on it, so this is a second view, not a measurement. AIFS 2, raw forecasts: AIFS's RMSE was lower in 4 of 8 tested comparisons, higher in 0, and could not be separated in 4. A further 22 differences are shown without a verdict. Each value is AIFS's RMSE minus the other model's, in per cent of the other model's. Minus is a lower AIFS error (better). | Variable | Against | Day 1 | Day 3 | Day 5 | Day 7 | Day 10 | Days | |---|---|---|---|---|---|---|---| | 500 hPa height | IFS | -2% (no verdict) | -8% (lower, tested) | -14% (lower, tested) | -7% (not separated) | 0% (not separated) | 171 | | 500 hPa height | GFS | -34% (no verdict) | -32% (lower, tested) | -29% (not separated) | -19% (lower, tested) | -11% (not separated) | 171 | | 2 m temperature | IFS | -35% (no verdict) | -29% (no verdict) | -21% (no verdict) | -15% (no verdict) | -7% (no verdict) | 171 | | 2 m temperature | GFS | -46% (no verdict) | -41% (no verdict) | -33% (no verdict) | -25% (no verdict) | -14% (no verdict) | 171 | | 10 m wind | IFS | -26% (no verdict) | -18% (no verdict) | -15% (no verdict) | -8% (no verdict) | -2% (no verdict) | 171 | | 10 m wind | GFS | -44% (no verdict) | -35% (no verdict) | -27% (no verdict) | -17% (no verdict) | -11% (no verdict) | 171 | Page: https://meteosignal.com/era5/ ## Licence and citation Results and text: CC BY 4.0, https://creativecommons.org/licenses/by/4.0/. Cite as: Soltanzadeh, I. (2026). MeteoSignal: verification of open weather forecasts over Europe. Data version 16 November 2026, run 20261116T194902Z. https://meteosignal.com/